Adaptive vibration assisted hybrid potential enhanced cascaded multi-stable stochastic resonance in bearing fault diagnosis

Stochastic resonance (SR) encounters significant limitations in noisy environments for bearing fault diagnosis applications. An Ensemble Empirical Mode Decomposition and Vibration Resonance-assisted Hybrid Potential Cascaded Multi-stable Stochastic Resonance (EEMD-VR-HPCMSR) system is proposed. The novel hybrid potential function combines an absolute value power term with three Gaussian functions, enabling adaptive transitions between mono-stable and multi-stable states, and uses Quantum Genetic Algorithm (QGA) to achieve independent optimization of multiple parameters. Combined with cascaded feedback system and vibration resonance, it greatly improves signal processing capability. Finally, in the fault detection experiment of the bearing dataset at the German PADERBORN and HUST, the performance of the system model was better than that of the classical bi-stable stochastic resonance (CBSR), classical tri-stable stochastic resonance (CTSR), and vibration resonance-assisted cascaded classical tri-stable stochastic resonance (EEMD-VR-CCTSR) systems. The results indicate that the system has superior performance in enhancing signal-to-noise ratio (SNR) and extracting fault features under strong noise conditions.